Towards a simplified taxonomy of Capsella bursa-pastoris (L.) Medik. (Brassicaceae) A. AKSOY', W. H. G. HALE" and J. M. DIXON Capsella bursa-pastoris (L.) Medik. is a species with a cosmopolitan distribution which shows considerable morphological variation. Numerous authors have recognised widely differing numbers of varieties, mi- crospecies or other infraspecific subdivisions (segregates) of this species. In an attempt to clarify this situation, we grew British material of the species under controlled conditions through to the F) generation to remove environmental variation, and assessed the plants on the basis of a range of morphological criteria, namely leaf shape, capsule size and also length of time taken to flower. Analysis of these characteristics consistently produced four basic groups, which had been previously described. Herbarium specimens could also nearly always be assigned to one of these groups. Limited chromosome counts suggest that two of these groups are diploid and two are tetraploid. We suggest this fourfold division into broad groups reflects the major genetic separations within the species, but that there is also considerable phenotypic plasticity shown by C. bursa- pastoris in response to factors such as shade or trampling. These four groups appear to differ in their geographic.al distribution in Britain. KEYWORDS:Shepherd's Purse, morphological variation, leaf characters, capsule characters, chromosome counts, infraspecific groupings. Capsella bursa-pastoris (L.) Medik. (Shepherd's Purse) (Brassicaceae) has a cosmopolitan distri- bution, and is a colonising species of disturbed ground. Being found in a broad range of conditions, up to 5900 m (Wilson 1949; Mani 1978) and in almost all countries of the world from tropical to subarctic habitats (Holm et al. 1979), the species is known to exhibit considerable morphological variation. Capsella bursa-pastoris has been described by numerous authors since the late 19th century, and has been divided taxonomically into many species, subspecies, varieties, microspecies and segre- gates. Jordan (1864), one of the earliest workers, described five species in France, namely Capsella agrestis, C. virgata, C. ruderalis, C. sabulosa and C. praecox, none of which are recognised today. Hopkirk (1869) considered the variation in Belgium to consist of subspecies derived from one common type, and he went on to describe six subspecies based primarily on the character of the capsule. Mott (1885) described eight varieties for Leicestershire and Rouy & Foucaud (1893) listed seven varieties and four subspecies based on the fruit characteristics in France. Almquist (1907) described 70 elementary species and later (Almquist 1921) examined British Capsella bursa- pastoris and listed 16 species. His descriptions were based on fine distinctions of leaf and capsule shape and size. Two years later, Almquist (1923) had recorded twelve classes of Capsella containing almost 200 microspecies. His microspecies were again based on minute differences in capsule shape and size, differing leaf shapes and position of leaf lobes. More recently, but only in Cyprus, Meikle (1977) recognized two species based on capsule size whilst Clapham et al. (1987) record Capsella bursa-pastoris as "very variable with a strong tendency for distinctive populations to arise because of self-pollination. Many of these have been named", but they do not specify any of these. The first edition of Flora Europaea (Chater 1964) comments that numerous variants have been described by Almquist, whilst the second edition (Chater 1993) states that "there is extreme * Present address: Department of Biology, Faculty of Science and Art, University of Erciyes, Kayseri, Turkey. ** Corresponding author A. AKSOY, W. H. G. HALE AND J. M. DIXON polymorphism within the four species listed" and that "Capsella bursa-pastoris is especially polymorphic and its variants incorporate many of the characteristics of the other three species". Most recently, Stace (1997) describes C. bursa-pastoris "as extremely variable in leaf and fruit shape; c. 25 segregates have been recognized in the British Isles". No details of these are given, nor is any information on them provided by the specialist CruCIfers of GrEat BrItain and IrEland (Rich 1991). The problems of taxonomy at the infraspecific level, relevant to a very variable species such as C. bursa-pastoris which is largely inbreeding yet has many phenotypic variations, are discussed, for example, by Stace (1989). This author notes that such phenotypic modifications would not be given taxonomic status by most taxonomists and that when such variations are recognised as phenotypic, they are relegated to synonomy. Capsella bursa-pastoris certainly shows phenotypic variation as a direct result of a wide range of environmental factors including temperature, shading, altitude, latitude and season; for example, Almquist (1923) found that leaves developing in autumn and spring were mostly lobed, whereas mid-summer leaves tend to be poorly lobed or entire. Hurka (1990) found pronounced ecotypic variation in time to flowering between early Scandinavian and late Alpine populations; he also found early and late ecotypes in North America. Aksoy (1996) observed the effects of shading on leaf shape, Steinmayer et al. (1985) recorded correlations between leaf form and temperature and rainfall, and Neuffer (1989) investigated the effects of temperature on variables such as leaf shape and flowering times. However, it is possible that not all the complex variations observed in this species can be reduced to phenotypic variation superim- posed on one single species complex. Shull (1909) collected seeds of C. bursa-pastoris from different sites in North America, where the species is introduced but now widely naturalized, and grew these under standard conditions for several generations, by self-pollination. He found that the majority of his plants could be fitted into four basic groups based on the characters of the rosette leaf shape. He referred to these four groupings ("biotypes" sensu Shull) as Capsella bursa-pastoris and used these names for them: rhomboidea, simplex, heteris and tenuis. Steinmayer et al. (1985) examined 29 populations of C. bursa-pastoris from the Alps to northern Scandinavia, from Iceland and also a population from the Hindu Kush Mountains in Afghanistan, while Neuffer (1989) worked with populations of C. bursA-pastoris collected from southern to northern Europe (including three from Britain), two populations from Egypt and one from Israel. Their populations were grown under standard conditions either in glasshouses or in field trials for varying periods of time. Analyses of leaf shapes and capsule size allowed the authors to classify most of their plants into one of the four basic groups proposed by Shull (1909). This paper seeks to expand on the work of Neuffer (1989) by examining populations of C. bursa-pastoris from a variety of habitats and geographical areas in Britain, to determine how well Shull's four basic groups are generally recognizable in the field, and from herbarium specimens; to germinate and grow seed from the different populations under standard conditions, and to observe whether, when environmental variation is removed, the plants can be classified according to Shull's groups; and, finally, to determine whether or not this classification is maintained in their progeny. If Shull's groupings are substantiated then a step will have been made towards simplifying the taxonomic classification of C. bursa-pastoris from 200 microspecies, or 25 segregates, or no attempt at all to sub-divide this variable species, to producing four useful groups, which can be recognized as having a distinct genetic basis underlying the environmentally modified phenotype. Although Shull's work was concerned only with examining leaf morphology, other workers, mentioned above, have used capsule size and shape as identifying characters and these have also been examined in the present paper; chromosome numbers have also been assessed. Chromosome numbers are usually tetraploid with 2n = 32 (Davis 1965; LOve & LOve 1956; Svensson 1983; Clapham et at. 1987). Chater (1993) records both 2n = 32 and also 2n = 16; Svensson (1983) also records diploid specimens with 2n = 16 from Greece. Seed samples from locally available populations were collected from 20 different habitats in Bradford and district, and from 14 other locations throughout Britain, between April and July 1993. These were germinated in potting compost in a glasshouse and grown on until they set seed. Cross pollination was assumed to be prevented by keeping each of the populations in a different place in the glasshouse, with sliding separating doors to aid isolation, and by the fact that the species is primarily adapted for self-pollination. Subsequent seed collection and growing was continued through to the F) generation. In each generation the rosette leaf shape was assessed for 15 plants, randomly selected from each population, according to a four-fold categorisation on the basis of Shull (1909) as follows: Capsella bursa-pastoris group A ("simplex" sensu Shull (1909» Leaves with mostly simple, rounded or triangular, acutish lobes. Capsella bursa-pastoris group B ("rhomboidea" sensu Shull (1909» Leaves divided to the midrib; possessing a more or less rhombic terminal lobe, set off by deep sinuses from the nearest lateral lobes. Capsella bursa-pastoris group C ("heteris" sensu Shull (1909» Leaves divided to the midrib; the terminal lobe usually